iscream

Reading WGBS data into analysis-ready structures

James Eapen, Jacob Morrison, Hui Shen

July 25, 2024

Whole genome bisulfite sequencing

Converting an epigenetic difference into a genetic difference

mC \(\xrightarrow{bisulfite}\) C

C \(\xrightarrow{bisulfite}\) U

Morrison et al. “Dupsifter” (2023, Bioinformatics)

DNA methylation analysis

WGBS alignment

Bismark
BISCUIT

Methylation analysis

BSseq: read.bismark \(\to\) bsseq \(\to\) DMRSeq, DMRcate
biscuiteer: readBiscuit \(\to\) BSseq
scMET

Krueger & Andrews “Bismark” (2011, Bioinformatics)
Zhou et al. “BISCUIT” (2024, Nucleic Acids Research)
Hansen et al. “BSmooth” (2012, Genome Biology)
Korthauer et al. “Detection and Accurate False Discovery Rate Control of Differentially Methylated Regions from Whole Genome Bisulfite Sequencing” (2019, Biostatistics)
Peters et al. “Calling Differentially Methylated Regions from Whole Genome Bisulphite Sequencing with DMRcate” (2021, Nucleic Acids Research)
Kapourani et al. “scMET” (2021, Genome Biology)

Analysis

BSseq

read.bismark to read Bismark .cov files
M and Coverage count matrices
GRanges storing CpG loci
Used by DMRSeq, DMRcate

An object of type 'BSseq' with
  69349 methylation loci
  93 samples
has not been smoothed
All assays are in-memory

read.bismark(files)

Hansen et al. “BSmooth” (2012, Genome Biology)

Analysis

BSseq

read.bismark to read Bismark .cov files
M and Coverage count matrices
GRanges storing CpG loci
Used by DMRSeq, DMRcate

An object of type 'BSseq' with
  69349 methylation loci
  93 samples
has not been smoothed
All assays are in-memory

gr <- GRanges([CpG loci])
read.bismark(files, gr)

Hansen et al. “BSmooth” (2012, Genome Biology)

iscream overview

Implementation

Querying and region operations written in Cpp
Queries made with htslib
Multithreaded across input files

Usage

data.frame <- region_map(
  tabixed_bedfiles_vec,
  genomic_interval_string_vec,
  fun = "aggregate", # or average
  bismark = FALSE,   # TRUE if biskmark .cov
  mval = TRUE        # FALSE for beta value
)

BSseq <- query_all(
  tabixed_bedfiles_vec,
  genomic_interval_string_vec,
  bismark = FALSE,
  merged = TRUE, # FALSE for uncollapsed strands
)

library(iscream)
options("iscream.threads" = 8)
files <- list.files(
  "aligner_output",
  pattern = "*.bed.gz",
  full.names = TRUE
)
regions <- (
  data.table::fread("windows.bed")
  [, paste0(V1, ":", V2, "-", V3)]
)
region_map(files, regions, fun = "aggregate")
query_all(files, regions)

Benchmarks

Runtime

CPU: Intel Xeon CPU E5-2698 v4 @ 2.20GHz

Benchmarks

Memory usage

CPU: Intel Xeon CPU E5-2698 v4 @ 2.20GHz

Benchmarks

Runtime

CPU: Intel Xeon CPU E5-2698 v4 @ 2.20GHz

Benchmarks

CPU: Intel Xeon CPU E5-2698 v4 @ 2.20GHz

Future directions

Fast reading of full files
Support for converting to and reading from arrow files and datasets
Comprehensive data checks
Sparse matrix support

Summary

iscream is an R package to read WGBS data into analysis-ready data structures
Can query genomic regions to extract CpGs only from regions of interest
Collect all CpGs into matrices or aggregate/average them within windows
Queries are parallelized across files

Acknowledgements

Shen lab

Ian Beddows
Svetlana Djirackor
Ben Johnson
Kelly Kryzanowski
Paige Matusiak
Amy Nuffesse (Admin)
Mary Rhodes
Ayush Semwal
Tashi Zandstra

github.com/huishenlab/iscream

References

Hansen, K. D., Langmead, B., & Irizarry, R. A. (2012). BSmooth: From whole genome bisulfite sequencing reads to differentially methylated regions. Genome Biology, 13(10), R83. https://doi.org/10.1186/gb-2012-13-10-r83

Kapourani, C.-A., Argelaguet, R., Sanguinetti, G., & Vallejos, C. A. (2021). scMET: Bayesian modeling of DNA methylation heterogeneity at single-cell resolution. Genome Biology, 22(1), 114. https://doi.org/10.1186/s13059-021-02329-8

Korthauer, K., Chakraborty, S., Benjamini, Y., & Irizarry, R. A. (2019). Detection and accurate false discovery rate control of differentially methylated regions from whole genome bisulfite sequencing. Biostatistics, 20(3), 367–383. https://doi.org/10.1093/biostatistics/kxy007

Krueger, F., & Andrews, S. R. (2011). Bismark: A flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics, 27(11), 1571–1572. https://doi.org/10.1093/bioinformatics/btr167

Morrison, J., Zhou, W., Johnson, B. K., & Shen, H. (2023). Dupsifter: A lightweight duplicate marking tool for whole genome bisulfite sequencing. Bioinformatics, 39(12), btad729. https://doi.org/10.1093/bioinformatics/btad729

Peters, T. J., Buckley, M. J., Chen, Y., Smyth, G. K., Goodnow, C. C., & Clark, S. J. (2021). Calling differentially methylated regions from whole genome bisulphite sequencing with DMRcate. Nucleic Acids Research, 49(19), e109–e109. https://doi.org/10.1093/nar/gkab637

Zhou, W., Johnson, B. K., Morrison, J., Beddows, I., Eapen, J., Katsman, E., Semwal, A., Habib, W. A., Heo, L., Laird, P. W., Berman, B. P., Triche, T. J., Jr., & Shen, H. (2024). BISCUIT: An efficient, standards-compliant tool suite for simultaneous genetic and epigenetic inference in bulk and single-cell studies. Nucleic Acids Research, gkae097. https://doi.org/10.1093/nar/gkae097